Cardiovascular Autonomic Neuropathy Is Associated With Increased Glucose Variability in People With Type 1 Diabetes

Cardiovascular autonomic neuropathy (CAN) is a common and severe complication of diabetes linked to increased mortality and morbidity (1), with a prevalence of 20–60% (2). Dysglycemia is an established risk factor for CAN in diabetes (3). Glucose variability (GV) may be associated with autonomic dysfunction in people with type 1 diabetes (T1D) (4). Conversely, autonomic dysfunction may induce adverse changes in glucose metabolism (5). Thus, the causal direction of detrimental GV fluctuations and CAN is unknown. The clinical CAN diagnosis has not been associated with GV. Demonstrating such an association would enable the identification of people with detrimental glucose excursions, which by intervention could be mitigated.

We aimed to investigate the possible association between the CAN diagnosis and GV in T1D.

Participants with T1D with established CAN or with no CAN were enrolled in a cross-sectional study from an ongoing neuropathy screening study at Steno Diabetes Center Copenhagen (Copenhagen, Denmark). Participants attended one physical study visit at which baseline characteristics were obtained. In this study, the CAN status was reaffirmed, and continuous glucose monitors (CGMs) were attached. Monitors and transmitters were returned at a second visit. The study was approved by the regional Committee (H-19029796) and the Danish Data Protection Agency (P-2019–291).

Forty-eight participants were included by the following criteria: 20–80 years of age, T1D (World Health Organization criteria), and having CAN or no CAN. Participants with or without CAN were matched on age, sex, and CGM use.

Participants were excluded in the presence of ischemic cardiovascular disease, thyroid disease, atrial fibrillation, pacemaker, alcohol abuse, insulin pump use, pregnancy, cancer treatment, hypoglycemia at first visit (capillary blood glucose <3.9 mmol/L), or use of β-blockers or antidepressants.

The CAN diagnosis was assessed by the three cardiovascular reflex tests (CARTs) recommended for diagnosing CAN (Vagus device; Medicus Engineering, Aarhus, Denmark) (6): change in heart rate during lying-to-standing test (30/15), deep breathing test (E/I), and Valsalva maneuver (VM). Tests were performed after at least 2 h of fasting and caffeine refrainment and 24-h abstinence from strenuous activity.

The age-dependent cutoff levels of CARTs (7) were used to diagnose CAN in the presence of at least two pathological CARTs (1). Participants with no pathological CARTs constituted the control group.

Masked 10-day CGM were recorded (Dexcom G6; Dexcom). If a minimum of 7 days were recorded, data were included (8). Participants continued routine CGM use if part of prestudy treatment. The glycemic metrics determined were: mean sensor glucose, coefficient of variation (CV), SD, mean amplitude of glycemic excursions (MAGE), continuous overlapping net glycemic action, time in range (TIR) 3.9–10 mmol/L, time above range (TAR) >10 mmol/L, and time below range (TBR) <3.9 mmol/L.

Distal symmetric polyneuropathy was assessed by the following and by use of age- and height-dependent thresholds where possible: light touch perception by 10-g monofilament, pain sensation by 40-g pinprick, vibration perception threshold, sural nerve function by conduction velocity, and amplitude potential as applied in previous studies (9).

The Michigan Neuropathy Screening Instrument was used to assess neuropathy symptoms, with a score ≥4 constituting neuropathy (10). Gastroparesis was assessed by the Gastroparesis Cardinal Symptom Index; a score >1.9 constituted gastroparesis. Gastroparesis cases were not confirmed by patient record screening. Hypoglycemia unawareness was evaluated by a validated questionnaire according to Pedersen-Bjergaard. Unawareness was categorized as: aware, impaired awareness, and unaware (11).

For descriptive statistics of continuous data, the Student t test was used for parametric data and Wilcoxon-Mann-Whitney test was used for nonparametric data. For categorical variables, a χ² test (if exact value ≥5) or Fisher exact test (if exact value <5) was used.

Linear regression analyses were applied to assess associations between the CAN diagnosis or CARTs as determinants and GV measures as outcomes by unadjusted models, as no relevant confounders were identified. Exploratory adjustments were done for HbA_1c and diabetes duration. Estimates are presented as percentages (95% CIs).

CARTs were log1.5-transformed. GV outcomes were log-transformed by the natural logarithm. Estimates were backtransformed. Estimates are difference in percentage in outcomes by a 50% lower CART. Analyses were performed in SAS Enterprise Guide 7.1 (SAS Institute).

The primary outcome was difference in CV between participants with and without CAN. We hypothesized a difference in CV between participants with and without CAN of 7, with an SD of 0.065 mmol/L (12). With 90% power and a two-sided significance level of 0.05, the sample size needed was 20 in each group. With dropout of ∼10%, the sample size was set at 24 per group.

Fifty-eight participants were screened: eight participants were excluded because of early CAN (one pathological CART) and two participants dropped out during CGM due to insufficient data (n = 1) or bleeding from the insertion site (n = 1). This left 48 participants with T1D for analyses. Twenty-four participants had CAN, and 24 participants had no CAN (mean age [years]: 54.6 ± 14.4 vs. 54.0 ± 12.8, P = 0.875; HbA_1c [mmol/mol]: 66.4 ± 12.5 vs. 57.8 ± 7.7, P = 0.006; HbA_1c [%]: 8.2 ± 1.1 vs. 7.4 ± 0.7, P = 0.006; diabetes duration [years]: 34 ± 14 vs. 23 ± 12, P = 0.004). Eighteen participants in each group used prestudy CGM; among these, eight (CAN group) and two (control subjects) had alarm features (Table 1). Four patients with CAN had gastroparesis. No cases were confirmed in their patient records.

Participants in the CAN group presented a higher degree on all measures of peripheral neuropathy when compared with control subjects (Table 1).

Participants with CAN had a higher CV, SD, and MAGE compared with participants with no CAN with higher estimates as follows: CV, 4.9% (95% CI 1.1, 8.7; P = 0.014); SD, 0.7 mmol/L (95% CI 0.3, 1.1; P = 0.002); and MAGE, 1.4% (95% CI 0.0, 2.7; P = 0.047). No other differences in GV indices or hypoglycemia awareness were found (Table 1). In models adjusted for diabetes duration and gastroparesis, CV and SD retained significant associations, while CV retained significant associations in HbA_1c-adjusted models (data not shown).

A 50% lower 30/15 ratio was associated with a higher CV and SD: 21% (95% CI 3, 43; P = 0.026) and 34% (95% CI 11, 63; P = 0.004), respectively. A 50% E/I ratio was associated with a higher CV and SD: 26% (95% CI 1, 57; P = 0.039) and 50% (95% CI 18, 92; P = 0.002), respectively. A 50% lower VM ratio was associated with a higher SD, lower TIR, and higher TAR: 21% (6, 39; P = 0.006), −25% (−42 to −3; P = 0.033), and 35% (2, 79; P = 0.034), respectively (Supplementary Table 1).

The main finding of the study was that CAN diagnosis in T1D was associated with higher CV, SD, and MAGE. This clear association between CAN and greater GV has not been demonstrated previously. In addition, we demonstrated that lower values of all CARTs were associated with greater GV. This indicates that both parasympathetic and sympathetic dysfunction are associated with increased GV, as the E/I ratio is mainly a parasympathetic measure, while the 30/15 ratio and VM ratio are mixed parasympathetic and sympathetic measures (13). Also, lower values of the VM ratio were associated with more time spent in hyperglycemia, indicating that autonomic dysfunction is attributable to the risk of time spent in detrimental hyperglycemia.

We saw no group differences in TIR, TAR, or mean sensor glucose despite a higher HbA_1c, indicating that the study may have induced better self-management behavior. Despite this bias, CAN was still associated with higher GV.

CAN reduces counterregulatory catecholamine responses, increasing the risk of severe hypoglycemia (14). However, no association between hypoglycemia (TBR) and the CAN diagnosis was seen. This could be explained by routine use of CGM with alarms (15). Eight of 18 CGM users with CAN had alarms. In contrast, 2 of 18 CGM users with no CAN had alarms. This imbalance could have biased the results.

The CAN diagnosis in T1D associated with higher GV. In addition, both parasympathetic and sympathetic dysfunction were associated with greater GV and with more time in hyperglycemia. Thus, CAN screening in T1D may enable identification of people with detrimental GV. Interventional studies are needed to investigate causality between CAN and GV.

Funding. The work leading to this article received funding from internal sources.

The funders of the study had no role in study design, data collection, analysis, interpretation, or writing of the report.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.H.N. was responsible for formal analysis, investigation, and writing the original draft of the article. H.I.M. was responsible for review and editing of the article. N.L. was responsible for investigating and review and editing of the article. J.C.L. was responsible for formal analysis and review and editing of the article. H.K. was responsible for review and editing of the article. M.M.B.C. was responsible for review and editing of the article. A.G.R. was responsible for formal analysis and review and editing of the article. K.N. was responsible for conceptualization, supervision, and review and editing of the article. C.S.H. was responsible for conceptualization, supervision, formal analysis, and review and editing of the article. C.S.H. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.